Structural and Functional Characterization of a Hole−HoleHomodimer Variant in a “Knob-Into-Hole” Bispecific AntibodyHui-Min Zhang,*,† Charlene Li,† Ming Lei,† Victor Lundin,† Ho Young Lee,‡ Milady Ninonuevo,†

Kevin Lin,§ Guanghui Han,∥ Wendy Sandoval,∥ Dongsheng Lei,⊥ Gang Ren,⊥ Jennifer Zhang,†

and Hongbin Liu*,†,○

†Protein Analytical Chemistry, ‡Biological Technologies, §Analytical Operations, and ∥Departments of Microchemistry, Proteomicsand Lipidomics, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States⊥The Molecular Foundry, Lawrence Berkeley National Laboratory, One Cyclotron Road, Berkeley, California 94720, United States

*S Supporting Information

ABSTRACT: Bispecific antibodies have great potential to be the next-generationbiotherapeutics due to their ability to simultaneously recognize two different targets.Compared to conventional monoclonal antibodies, knob-into-hole bispecific antibodiesface unique challenges in production and characterization due to the increase in variantpossibilities, such as homodimerization in covalent and noncovalent forms. In this study, astorage- and pH-sensitive hydrophobic interaction chromatography (HIC) profile changewas observed for the hole−hole homodimer, and the multiple HIC peaks were exploredand shown to be conformational isomers. We combined traditional analytical methodswith hydrogen/deuterium exchange mass spectrometry (HDX MS), native massspectrometry, and negative-staining electron microscopy to comprehensively characterizethe hole−hole homodimer. HDX MS revealed conformational changes at the resolutionof a few amino acids overlapping the CH2-CH3 domain interface. Conformationalheterogeneity was also assessed by HDX MS isotopic distribution. The hole−holehomodimer was demonstrated to adopt a more homogeneous conformational distributionduring storage. This conformational change is likely caused by a lack of CH3 domain dimerization (due to the three “hole” pointmutations), resulting in a unique storage- and pH-dependent conformational destabilization and refolding of the hole−holehomodimer Fc. Compared with the hole−hole homodimer under different storage conditions, the bispecific heterodimer, guidedby the knob-into-hole assembly, proved to be a stable conformation with homogeneous distribution, confirming its high qualityas a desired therapeutic. Functional studies by antigen binding and neonatal Fc receptor (FcRn) binding correlated very well withthe structural characterization. Comprehensive interpretation of the results has provided a better understanding of both thehomodimer variant and the bispecific molecule.

Bispecific antibodies (BsAbs) have gained significantinterest in the biotech field as therapeutic alternatives to

conventional monoclonal antibodies (mAbs).1−3 Currently,more than 60 BsAb formats exist,1,3 and they can bedifferentiated into two major classes: with or without an Fcregion.4 The BsAbs with an Fc region are more popular becauseof their known stability and long half-life due to their large sizeand neonatal Fc receptor (FcRn)-mediated recycling process.4,5

The fermentation and purification processes established for theproduction of standard IgG molecules can often be leveragedfor such BsAbs. Different from conventional mAbs, which havetwo identical antigen-recognizing or Fab moieties, BsAbs havetwo different paratopes on the variable domains recognizingtwo different antigens. This unique feature results in morecomplex variants for BsAbs, for example, homodimers from thetwo different arms, mispairing of light chains, and so on. Toproduce BsAbs, the “knob-into-hole” format has been adoptedto promote heavy-chain heterodimerization of the two halfantibodies.6,7 The knob (T366W) and hole (T366S, L368A,and Y407 V) mutants are located at the CH3 domain interface,

and thus should not impact antigen binding or Fc function. Thelight chain mispairing problem can be overcome duringproduction by several approaches.3 One approach is throughan in vitro assembly step, where two half antibodies areexpressed in two different host cells. After two separate ProteinA affinity capture steps, the two half antibodies are mixed for invitro assembly by reduction and oxidation8 followed bydownstream purification of the BsAb.The knob-into-hole design and in vitro assembly can

efficiently drive heterodimerization of the heavy chains;however, some level of homodimers (including knob−knobhomodimers and hole−hole homodimers) are still presentduring half antibody purification and BsAb assembly.7 In fact,homodimers are the typical minor forms in the affinity-capturedpools of each arm. It is challenging to separate the homodimers

Received: September 19, 2017Accepted: November 12, 2017Published: November 12, 2017

Article

pubs.acs.org/acCite This: Anal. Chem. 2017, 89, 13494−13501

© 2017 American Chemical Society 13494 DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

from the BsAbs and quantify their levels, depending on thephysicochemical property differences between the two arms.The homodimer variants also have different forms, for example,covalent or noncovalent forms.9 Elliott et al. discovered themolecular details for knob-into-hole and homodimer inter-actions in the Fc of an IgG1 BsAb.10 They solved X-ray crystalstructures of both the knob−knob and hole−hole Fc fragmenthomodimers, which revealed an antiparallel Fc orientation.Intact mass analysis has also been used for bispecific variantsidentification and quantitation.9,11,12

Homodimers can potentially cause undesired bioactivity. Inaddition, because of their potentially unnatural structure andpresumed low stability, the homodimers may be more prone toaggregate which may lead to increased immunogenicity.11

Detailed characterization and close monitoring of thehomodimers during the production process are crucial todevelop safe and efficacious BsAbs.When typical biophysical and analytical methods used for

conventional mAbs are applied to characterize BsAbs, it is verychallenging to successfully separate and monitor the homo-dimers. There is increasing interest in using high-resolutionstructural methods to characterize mAbs. As is typical for IgG,there is no full-length BsAb X-ray structure available, and onlythe Fc of a BsAb has been solved.10,13 By monitoring thedynamics of backbone amide hydrogen exchange intodeuterium, hydrogen−deuterium exchange mass spectrometry(HDX MS) can probe protein dynamics and conformationalchange in solution. In recent years, HDX MS has been widelyused to characterize protein conformation and protein−proteininteractions,14−17 for example, epitope mapping,18 conforma-tional change due to chemical modifications,19,20 or aggregationmechanism analysis.21,22 The interaction of IgG with itspurification resin protein A and the pH-dependent interactionwith FcRn have also been reported.22,23 Pan et al. discussed theconformational impact of drug conjugation on the mAb.24

In this study, a slow change during storage at 2−8 °C wasobserved by hydrophobic interaction chromatography (HIC)for the hole−hole homodimer variant of a knob-into-hole BsAb(BsAb1). No chemical modifications were detected duringstorage, indicating that the HIC peaks are multiple conforma-tional isomers for this variant. The different conformationalisoforms and their changing profile during storage are ofconcern not only for its quantitation and control, but also for itspotential bioactivity and immunogenicity risks in a therapeuticcontext. We combined traditional methods, such as HIC, sizeexclusion chromatography (SEC), and liquid chromatography−mass spectrometry (LC-MS), with state-of-the-art methods, forexample, native MS, HDX MS, and electron microscopy(EM)25 to characterize the conformations of the knob-into-holeBsAb and the hole−hole homodimer during storage. Theconformational changes identified by HDX MS were mappedto an IgG1 structural model for the hole−hole homodimervariant, and the results were consistent with all otherexperimental observations, including antigen binding andFcRn binding data. To our knowledge, this is the firstdemonstration of HDX’s ability to detect the structural changeof a molecule without any primary structural change (chemicalmodifications) or matrix change, since no binding partners werepresent and pH stayed constant at around 5.8 during storage at2−8 °C. This approach has helped us better understand theroot cause of its structural change and better control thisunique type of BsAb impurity.

■ EXPERIMENTAL SECTION

Production of the Hole−Hole Homodimer for BsAb1.The hole−hole homodimer of BsAb1, which is an in vitroassembled knob-into-hole BsAb, was produced and purified atGenentech Inc. (South San Francisco, CA). The harvested cellculture fluid of the hole half antibody was purified throughprotein-A affinity chromatography. The pH of the protein-Apool was adjusted from 3.3 to 5.0, and Poros cation exchangechromatography (CEX) was used to separate the hole−holehomodimer from the half antibody and other species. Theelution pH was 5.5. Ultrafiltration/diafiltration was applied andconditioned at pH 5.8 in 20 mM histidine acetate formulationbuffer. The resulting hole−hole homodimer was kept at −70°C for storage. The identity of the isolated hole−holehomodimer and the knob-into-hole bispecific heterodimerwere confirmed by intact mass analysis, which also confirmedthat the hole−hole homodimer was a covalent homodimer.

Hydrophobic Interaction Chromatography. HIC wasperformed on a Waters Alliance 2695 HPLC (WatersCorporation, Millford, MA) with a Dionex ProPac HIC-10column, 4.6 × 100 mm (Thermo Scientific, Waltham, MA).Mobile phase A was 20 mM Tris buffer (pH 7.5) containing 1.5M ammonium sulfate. Mobile phase B was 20 mM Tris buffer(pH 7.5). Each sample was diluted to 2 mg/mL with LC-gradewater. The protein load for each injection was 40 μg. A lineargradient from 0 to 100% mobile phase B in 60 min was used.The column temperature was maintained at 25 °C, and the flowrate was 0.8 mL/min. The column effluent was monitored at280 nm.

Intact Mass Analysis by Native Mass Spectrometry.Native MS analysis with an Exactive Plus extended mass range(EMR) Orbitrap mass spectrometer was used to study thecharge-state distribution of the hole−hole homodimers. A briefdescription of the method is provided in the SupportingInformation.

Hydrogen/Deuterium Exchange Mass Spectrometry.The HDX MS experiments of hole−hole homodimers stored at−70 °C and 2−8 °C for 9 months were performed on a fullyautomated Leap robotic system (Leap Technologies, Carr, NC)connected to an Orbitrap Elite mass spectrometer (ThermoScientific). The two samples were at 5 mg/mL in 20 mMsodium acetate buffer (pH 5.3 ± 0.3) in H2O. The samesodium acetate buffer was also prepared in D2O for deuteriumlabeling, pD 5.3 ± 0.3.For deuterium labeling, 3.5 μL of the hole−hole homodimer

sample was mixed with 55 μL of the labeling buffer in D2Ostored at both −70 °C and 2−8 °C. The samples were labeledfor 30 s, 1 min, 10 min, 1 h, and 4 h in D2O buffers at 20 °C intriplicate, and quenched with 55 μL of quench solution (8 Murea, 1 M TCEP·HCl, pH 2.2). Immobilized protease XIII/pepsin (1:1) column (2.0 × 30 mm; NovaBioAssays, Inc.,Woburn, MA) was used for online digestion in 0.1% formicacid and 0.04% TFA in H2O (pH 2.3), at 100 μL/min. Thedigests were captured on a trapping column (WatersACQUITY BEH C18 VanGuard Precolumn 2.1 × 5 mm)and then eluted to a Waters BEH C18 UHPLC column (2.1 ×50 mm) for peptide separation on a Waters Nano AcquityHPLC system. A 12 min gradient of 5−50% B (A/B: 0.1%formic acid and 0.05% TFA in H2O/acetonitrile) was used, at50 μL/min. The eluted peptides were directed into an Orbitrapmass spectrometer with electrospray ionization for detection, inthe m/z range 300−1800 at a resolving power of 60000 at m/z

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13495

400. Peptides were identified using a combination of exact massand MS/MS aided by Mascot search. Peptide deuterium levelswere determined using EXMS26 and a python script.27 Theaverage relative deuterium uptake difference (ARDD)18 for allthe time points was also automatically calculated by themodified python script to represent the overall protection oneach peptide.Antibody Structure by Electron Microscopy and

Antigen Binding. Negative-staining electron microscopy(NS EM) using an optimized negative-staining (OpNS)protocol was used to study the structure of the antibodyhomodimers. Hole-hole homodimer binding with the antigenpeptide was measured using an antigen binding assay. Briefdescriptions of these methods are provided in the SupportingInformation.

■ RESULTS AND DISCUSSIONDifferent Conformational Isoforms Detected for the

Hole−Hole Homodimer. In an attempt to monitor andquantify the knob−knob and hole−hole homodimers in theknob-into-hole bispecific product, a HIC method wasdeveloped. Unlike the bispecific heterodimer, which has asingle HIC peak, the hole−hole homodimer displayed threemajor peaks when analyzed after being thawed from −70 °C(Figure 1A).

Intact mass analysis by a chip-TOF technique9 for theisolated HIC peaks revealed protein species with the samemasses, suggesting that the multiple HIC peaks representedconformational isomers. Peptide mapping analysis did notidentify any significant increase in chemical modifications, withonly a trace level (<0.5%) of Met 254 and Met 430 oxidationdetected in the −70 °C or the 2−8 °C sample. The lack ofdissociation between heavy chains during chip-TOF-MSanalysis (under denaturing condition) suggests that thehomodimer in question is covalently linked. The discovery ofmultiple conformations of the hole−hole homodimer wassurprising and prompted further study upon storage.

The HIC profile of the hole−hole homodimer had shifted tothe left upon storage at 2−8 °C for 1 month at pH 5.8 (Figure1A), indicating that the hydrophobicity slowly decreases. Thisconformational change was slow, because after 24 h at 2−8 °C,the profile showed little change. Subsequent experimentsrevealed that the HIC profile was also pH-dependent. Figure1B shows that immediately after the hole−hole homodimer wasbuffer exchanged from pH 5.8 to pH 3.0, it appears as a singlepeak eluting at the far right side of the chromatogram (morehydrophobic). However, for the same pH 5.8 starting material,after buffer exchange to pH 9.0, the HIC profile shifts to the left(less hydrophobic). Similar to storage at 2−8 °C, the basic pHcaused a decrease in the hydrophobicity of the sample. TheBsAb1, however, did not show any such storage- or pH-dependent change on the HIC profile; instead, it stayed as asingle peak eluting at the same position (data not shown).To confirm the slow profile change from high to low

hydrophobicity, the hole−hole homodimer was bufferexchanged to pH 3.0 and then to pH 7.5, followed by HICanalysis at different time points after storage at 2−8 °C. ThepH change from pH 3.0 to pH 7.5 caused a peak with lowerhydrophobicity on the left side of the HIC profile to increase(see Supporting Information, Figure S-1). However, even after18 h at 2−8 °C, the majority of the molecules stayed on theright side and only a small portion of the homodimer shifted tothe left, demonstrating that the conformational change is slow.SEC analysis was also performed, as described in the

Supporting Information, to assess the hydrodynamic radius(Rh) of the hole−hole homodimer samples. The early elutingspecies have a larger Rh than the late-eluting species, whichindirectly relates to the conformational change of the sample.Figure S-2 (see Supporting Information) displays the SEC datafor the control sample (stored at −70 °C) and the samplesstored at 2−8 °C for 2 weeks and 6 months. All three sampleseluted as monomer, indicating that the multiple speciesobserved in the HIC profile (Figure 1) represent thehomodimer monomer, not high molecular weight aggregate.The sample stored at −70 °C displayed two species shown as

two split peaks (Figure S-2), present in about equal abundance.The two-week sample has a higher abundance of the speciesthat elutes later, and the six-month sample has mainly onespecies that elutes later. The right-shifting SEC profile clearlyindicates that the Rh of the hole−hole homodimer continues todecrease after storing at 2−8 °C from 2 weeks to 6 months.Since the molecular weight stays the same, the Rh change canperhaps only be explained by conformational change, indicatingthat upon storage at 2−8 °C, the protein begins to take on amore compact conformation.

Structural Basis for the Conformational Isoforms fromNative MS. Since these HIC-separated peaks were most likelyconformational isomers of the hole−hole homodimer, wewanted to further characterize their structures. The charge-statedistribution by native MS can be related to the folding of aprotein, where a lower charge distribution generally means amore folded and compact structure.28 Samples in native buffer(in 100 mM ammonium acetate, pH 6.8) were infused onto amass spectrometer to compare the conformational difference ofthe hole−hole homodimers stored at the two conditions(Figure 2, upper panel, −70 °C; lower panel, 2−8 °C for 6months). Compared with the sample stored at −70 °C, thesample stored at 2−8 °C has a higher m/z distribution andlower charge states, indicating less solvent-exposed surface area.Clearly, the −70 °C sample displayed two Gaussian

Figure 1. HIC profile of BsAb1 and the hole−hole homodimer (HD)(A) at different storage conditions; (B) hole−hole HD frozen control,buffer exchanged to pH 3.0 and pH 9.0, respectively. Note: The −70°C control materials used in this figure were from different batches andformulated in different buffer conditions including pH [pH 5.8 for (A)and pH 5.5 for (B)]. Their HIC profiles served as the internalreference within the same experiment only.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13496

distributions, indicating two conformational species. The 2−8°C sample has m/z peaks (3800−4600 range) with twice thespacing as in the −70 °C sample, which corresponded to halfantibody masses due to the direct infusion of the samples (noseparation between the homodimer and the half antibody). Acloser look at the charge states for the −70 °C sample revealeda mixture of homodimer and half antibody, because the relativepeak abundance of the 36+ and 34+ is greater as compared witha regular Gaussian distribution, indicating two species in thesample.

Hole−Hole Homodimer Becomes More Compact afterStorage at 2−8 °C as Confirmed by HDX MS. To obtainhigh-resolution conformational information by HDX MS, it isessential to get high sequence coverage on the proteins beingcompared. Protease type XIII is superior to pepsin in terms ofgenerating more overlapping peptides in solution.29−31 For thisstudy, the column with protease type XIII and pepsinimmobilized at 1:1 ratio resulted in the highest sequencecoverage and the most number of unique peptides. For thehole−hole homodimer samples stored at −70 °C and 2−8 °C,we acquired 330 unique peptides and achieved high sequencecoverage at 94%.Under exactly the same experimental conditions, deuterium

uptake was compared for the hole−hole homodimer samplesstored at −70 °C and 2−8 °C for 9 months. As shown in Figure3A, the peptides in the Fab regions on both the heavy and lightchains did not show any difference in deuterium uptake (e.g.,

Figure 2. Native MS data for the hole−hole homodimer (HD) storedfor 6 months at (A) −70 °C and (B) 2−8 °C.

Figure 3. (A) Representative time-course HDX MS data for the hole−hole homodimer (HD) 2−8 °C vs −70 °C; (B) HDX MS results mapped ona human IgG1 mAb, 1HZH; light brown, light chain; gray, heavy chain; blue, large protection (ARDD ≤ −10%); and light blue, a small protection(−10% < ARDD← 3%); (C) Zoomed-in color-coded HDX MS results on the Fc region of a human IgG1 mAb: 1HZH. The peptides shown in (A)are labeled; (D) Proposed model for the hole−hole HD under different pH and storage conditions. After storage at 2−8 °C or pH increases, thehole−hole HD can slowly form a stable conformation by the Fc dimerization.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13497

HC CDR2 46−70), indicating no conformational difference inthe Fab region.Large regions on the heavy chain Fc region (CH2 and CH3)

displayed reduced deuterium uptake in the sample stored at 2−8 °C for 9 months, implying increased structural stability afterstorage at 2−8 °C. The HDX results are color-coded on aconventional IgG1 molecule (PDB ID: 1HZH, Figure 3B),similar to the hole−hole homodimer. The cold colors (blue andcyan) indicate more protection or less dynamics in the 2−8 °Csample compared with the −70 °C sample. A closer look at theFc region revealed details of the conformational change after2−8 °C storage (Figure 3C). On the CH2 domain, the largeprotection is located mainly on peptide a (HC 243−253). Onthe CH3 domain, the hole mutants T368S and L370A on ornear peptide c (HC 370−380), as well as Y409 V on peptide d(HC 407−416) all become protected after 2−8 °C storage.(Note: the numbering is off by two amino acid residues fromthe generic framework numbering reported in the literature.7)These two β-sheets where peptides c and d reside are normallyinvolved in the CH3 dimer interaction.6 Peptide e (HC 433−441), with two histidine residues (His435/His437), is alsolargely protected after storage at 2−8 °C. These high-resolutionHDX MS data demonstrated that some regions of the hole−hole homodimer Fc domain slowly fold together and form amore compact structure. These data are consistent with theresults showing that after storage at 2−8 °C, the homodimerbecomes less hydrophobic by HIC, has smaller Rh by SEC, andbears fewer charges by native MS.Potential Mechanism for the Conformational Change

in the Hole−Hole Homodimer. Since the hole−holehomodimer conformational change is impacted by storageconditions, it is important to understand if the conformation isaffected by the manufacturing process. After protein A affinitychromatography, the protein is eluted in pH 3.3 before

adjusting to pH 5.0, followed by CEX purification and bufferexchange to pH 5.8. It was previously reported that IgG1antibodies show reduced structural stability in the Fc domain atlow pH of 5.5 due to columbic repulsion from positivelycharged histidine residues at the junction between the CH2 andCH3 domains.23 It is proposed that under low pH, the proteinA-eluting conditions, the hole−hole homodimer (also an IgG1molecule) is unstable, and the CH2 and CH3 domains aredisconnected due to columbic repulsion from His312 in CH2and His435/His437 in CH3. The conformational change to amore hydrophobic and less-folded state with low pH (3.0)exposure (Figure 1B) supports this model. When the pH of thesolution is increased to 5.8 or higher, most of the histidineresidues return to neutral charge, and the columbic repulsionassociated with the Fc destabilization at low pH is greatlyreduced. However, the refolding of the hole−hole homodimerFc is slow due to the lack of a stable CH3-CH3 domain dimer(as a result of the three “hole” mutations). After prolongedstorage at pH 5.8 or higher, the Fc eventually reaches a localminimum energy state, as suggested by the reduced dynamicson both CH2 and CH3 domains by HDX MS.

Evidence from Negative-Staining Electron Micros-copy Data. To confirm the structural variation at themolecular imaging level, electron microscopy (EM) was usedto examine the sample of hole−hole homodimer after storage at−70 °C, as an orthogonal method to validate the proposedstructure based on HDX data. Representative NS EM particles’images showed two main morphology categories observed forthe hole−hole homodimer, Y- and X-shaped species (Figure 4).To confirm that the observed structures were statisticallysignificant, reference-free class averaging analysis was per-formed. The selected reference-free class averages confirmedthe existence of the Y- and X-shaped species. The Y-shapedspecies are similar to what is expected for a conventional IgG

Figure 4. NS EM of the hole−hole homodimer sample stored at −70 °C. Representative single-molecule EM raw images and selected reference-freeclass averages of (A) the Y- and (B) X-shaped molecules are indicated, with possible domain identification labeled at the bottom. Top row, the rawimages of particles and raw class averages; second row, their corresponding images after soft-boundary masking; third row, the models overlappedwith the masked images; bottom row, the structure models; scale bars = 10 nm.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13498

molecule. However, the X-shaped species was only observed inthis hole−hole homodimer sample, but not in conventional IgGor bispecific molecules. The X- and Y-shaped species differ inthe Fc region where the two halves are away from each other,implying less interaction between the two halves. The particleimages observed by EM are snap-shots of the dynamics of themolecule. Uranyl formate used in OpNS works at a lower pHvalue (∼4.5), which may partially denature the hole−holehomodimer’s conformation, as shown on Figure 3D. Because ofthe low pH (∼4.5) of the fixing reagent, a comparison of thehole−hole homodimer samples at different storage conditions,as shown in Figure 3, by HDX MS was not possible by EM,mainly due to the quick conformational change of the hole−hole homodimer at low pH.Knob-Into-Hole Bispecific Molecule Is the Most Stable

Form. The conformation of the desired knob-into-hole BsAb1was also compared with the hole−hole homodimers by HDX.Three samples (BsAb1 and the hole−hole homodimers at −70°C and 2−8 °C) were compared (Figure 5A). The BsAb1exhibits the least deuterium uptake in the Fc region, and it istherefore the most folded and stable form among the threesamples. Not surprisingly, the CH3 domain displayed thehighest stability in the BsAb1, consistent with the intendedpurpose of the knob and hole point mutations to promoteknob-into-hole side chain packing. Since the BsAb1 is morestable than the hole−hole homodimers, even after prolongedstorage at 2−8C, this implies an inability on the part of thehomodimer to form a native Fc conformation.As expected, no deuterium uptake difference was seen in the

complementarity determining regions (CDRs) comparingBsAb1 and the hole−hole homodimers, suggesting little orno impact due to the knob or hole mutations and/or Fcconformation. This high-resolution structural informationstrongly supports the knob-into-hole bispecific design that theFab of each half antibody is maintained, without any impactfrom the other half.EX1 Kinetics Observed for the Hole−Hole Homo-

dimer Conformational Change. In addition to theprocessed deuterium incorporation data, HDX MS raw dataalso provide insight into protein conformation. Two exchangekinetics of HDX MS have been described, EX2 and EX1.29,32,33

The EX2 exchange pattern usually represents a stable andhomogeneous protein population. The EX1 exchange kineticsusually implies a heterogeneous protein population withdifferent protein conformations. In Figure 5B, a CH2 peptide(243−253) displayed EX1 exchange pattern for the hole−holehomodimer stored at −70 °C; however, the sample stored at2−8 °C and the BsAb1 followed EX2 exchange kinetics. Inaddition, other peptides in the CH2 and CH3 domains that alsoexhibited EX1 distribution in the −70 °C hole−holehomodimer include: 309−320, 336−358, 407−428, and 428−448. Therefore, the hole−hole homodimer at −70 °C is amixture of different populations with different conformations.These conformations can be correlated with the different HICprofiles. The fast exchanging population is likely to be the morehydrophobic peak, and the slow exchanging population is likelyto be the least hydrophobic peak. After prolonged storage (9month) at 2−8 °C, the EX2 pattern suggests that thehomodimer becomes one homogeneous species shown as asingle low-hydrophobicity peak on the HIC profile (Figure 1A),again implying that it is adopting a stable and homogeneousconformation.

Antigen Binding and FcRn Affinity Chromatographyof Hole−Hole Homodimer Compared with BsAb1. Whencharacterizing a product-related variant of a biotherapeuticmolecule, it is important to understand whether it impacts thefunction of the molecule. For the BsAb1 molecule, a bindingassay was developed (see Supporting Information) to evaluatethe antigen binding activity of the hole arm of the BsAb andhole−hole homodimer. Here, this binding assay was used tostudy the impact of different hole−hole homodimer structuralisomers on the antigen binding activity (Figure 6). Since onlyone arm of the BsAb1 binds to the hole antigen (monovalentbinding), BsAb1 should show about 50% binding relative to thehomodimers (bivalent). The observed result (53% for BsAb1)demonstrated the capability of this assay to detect relevantdifferences in hole antigen binding activity. These results

Figure 5. (A) HDX MS time-course results for the hole−holehomodimer (HD) stored at 2−8 °C vs −70 °C compared with BsAb1;(B) Isotopic distribution of a CH2 peptide 243−253 after differentperiods of deuterium exchange. The −70 °C sample displays EX1exchange pattern, and the 2−8 °C and BsAb1 samples show EX2exchange pattern.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13499

indicate that after storage at 2−8 °C for 5 months, or at pH 3.0,even though their HIC profiles show dramatic change, there islittle or no difference in antigen binding for the conformationalisomers of the hole−hole homodimer. These binding data areconsistent with the HDX data showing no changes in Fabstructure (Figure 3B).An FcRn affinity chromatography method was used to

evaluate the interaction between FcRn and IgG Fc in vitro,providing insight into the structure that may affectpharmacokinetics (PK) in vivo.34 As shown in Figure S-3, theBsAb1 displayed a typical FcRn affinity chromatography profilewith a single main peak. In contrast, the −70 °C hole−holehomodimer sample showed multiple peaks, one with no FcRnbinding affinity, one with low-affinity FcRn binding and onewith normal FcRn affinity, whereas the hole−hole homodimerstored at 2−8 °C displayed a main peak and a small peak withlow-affinity binding. These results are consistent with the HICand HDX MS data and correlate well with the structural modelshown in Figure 3D. FcRn binds at the interface between theCH2 and CH3 domains, and the unfolded Fc (as shown on theleft and the middle of the proposed model in Figure 3D at −70°C) impairs FcRn binding. After storage at 2−8 °C, the Fcconformation changes so as to restore the FcRn binding ability(as shown on the right side of the proposed model in Figure3D), resulting in a single main peak profile (Figure S-3,middle).

■ CONCLUSIONSIn this study, we characterized the hole−hole homodimervariant of an in vitro assembled knob-into-hole bispecificantibody by using a suite of analytical and biophysical methods.Multiple conformational isomers were observed in the hole−hole homodimer of BsAb1 by HIC. The more hydrophobicisoforms, as shown by HIC, probably result from partial proteinunfolding during the low-pH production process, and theseisoforms slowly refold into more compact, less hydrophobicconformation during storage at pH 5.8 and 2−8 °C. Highresolution HDX MS identified the location of the conforma-tional change to be at the junction between the CH2 and CH3domains of the hole−hole homodimer. Based on the HDX MSresults, we propose that at −70 °C and pH 5.8, the Fc of thehole−hole homodimer is split, forming an X-shaped con-formation, as shown by the EM data. After storage at 2−8 °C,the Fc slowly refolds. From the isotopic distribution of theHDX MS data, we also discovered that the hole−holehomodimer stored at −70 °C exhibits heterogeneous

conformations, and after storage at 2−8 °C, a homogeneousconformation is generated.The cause of this unusual conformational change of the

hole−hole homodimer during storage at 2−8 °C was discussed.Since the HIC profile displayed reduced hydrophobicity afterstorage at 2−8 °C, we propose the conformational changeduring storage is due to the original pH-induced change duringpurification of the hole−hole homodimer. After it was purifiedfrom protein-A chromatography at pH 3.3, an increase to pH5.8 allowed the hole−hole homodimer to refold from a partiallydenatured conformation to a more stable and compactconformation very slowly (in the range of weeks when storedat 2−8 °C). This slow conformational change is likely causedby the hole mutations, resulting in disfavored CH3 dimerizationand the slow Fc dimerization/folding.The bispecific molecule, BsAb1, has the most stable

conformation with a homogeneous population, which confirmsthat the BsAb product are more stable in the purification andstorage processes due to the stabilization guided by the knob-into-hole design. We also confirmed that the antigen bindingactivity of the hole arm of the BsAb and hole−hole homodimerantibody were unchanged, and likewise confirmed that theconformation in the CDRs of the Fab regions were identical byHDX MS. The FcRn affinity chromatography results alsoconfirmed that after storage at 2−8 °C, the Fc of the hole−holehomodimer becomes more folded for better FcRn binding.The hole−hole homodimer variant is expected to have

undesired bioactivity. In addition, the unstable structure of thehole−hole homodimer during storage could pose potentialsafety or immunogenicity risk to patients. This detailedcharacterization helps to better understand the product-relatedhole−hole homodimer variant which could facilitate bettermonitoring of the variant during production, and ultimately aidin designing safer and more efficacious therapeutic antibodies.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.anal-chem.7b03830.

Methods used for SEC, intact mass analysis by nativemass spectrometry, NS EM, the antigen binding assay,and the FcRn affinity chromatography; Figure S-1: HICprofile of the hole−hole homodimers; Figure S-2: SECfor the hole homodimer stored under differentconditions and lengths of time; and Figure S-3: FcRnaffinity chromatograph for the hole−hole homodimerstored under different conditions and lengths of time(PDF).

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: zhang.huimin@gene.com. Tel.: 650-467-9618. Fax:650-225-3554.*E-mail: hliu@nektar.com. Tel.: 415-482-5423. Fax: 415-339-5381.ORCIDHui-Min Zhang: 0000-0002-2099-4730Ming Lei: 0000-0002-1247-1358Present Address○Nektar Therapeutics, 455 Mission Bay Blvd. South, SanFrancisco, CA 94158, United States.

Figure 6. Relative antigen binding of the hole−hole homodimers(HD) in the two storage conditions, −70 °C, 2−8 °C for 5 monthsand low pH (3.0). BsAb binding was run as a negative control.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13500

Author ContributionsAll authors have given approval to the final version of themanuscript.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors would like to thank Ben Walters, Judy Shimoni,and Feng Yang for their critical review of the manuscript; andYung-Hsiang Kao, John Stults, and John Joly for their supportof this project. Work at the Molecular Foundry was supportedby the National Science Foundation under Grant DMR-1344290, and the Office of Science, Office of Basic EnergySciences of the U.S. Department of Energy under Contract No.DE-AC02-05CH11231. D.L. and G.R. were partially supportedby the National Heart, Lung, and Blood Institute of theNational Institutes of Health (no. R01HL115153) and theNational Institute of General Medical Sciences of the NationalInstitutes of Health (no. R01GM104427).

■ REFERENCES(1) Spiess, C.; Zhai, Q.; Carter, P. J. Mol. Immunol. 2015, 67, 95−106.(2) Byrne, H.; Conroy, P. J.; Whisstock, J. C.; O’Kennedy, R. J.Trends Biotechnol. 2013, 31, 621−632.(3) Fan, G.; Wang, Z.; Hao, M.; Li, J. J. Hematol. Oncol. 2015, 8, 130.(4) Kontermann, R. E.; Brinkmann, U. Drug Discovery Today 2015,20, 838−847.(5) Kontermann, R. E. BioDrugs 2009, 23, 93−109.(6) Ridgway, J. B.; Presta, L. G.; Carter, P. Protein Eng., Des. Sel. 1996,9, 617−621.(7) Atwell, S.; Ridgway, J. B.; Wells, J. A.; Carter, P. J. Mol. Biol. 1997,270, 26−35.(8) Spiess, C.; Merchant, M.; Huang, A.; Zheng, Z.; Yang, N. Y.;Peng, J.; Ellerman, D.; Shatz, W.; Reilly, D.; Yansura, D. G.; Scheer, J.M. Nat. Biotechnol. 2013, 31, 753−758.(9) Macchi, F. D.; Yang, F.; Li, C.; Wang, C.; Dang, A. N.; Marhoul,J. C.; Zhang, H. M.; Tully, T.; Liu, H.; Yu, X. C.; Michels, D. A. Anal.Chem. 2015, 87, 10475−10482.(10) Elliott, J. M.; Ultsch, M.; Lee, J.; Tong, R.; Takeda, K.; Spiess,C.; Eigenbrot, C.; Scheer, J. M. J. Mol. Biol. 2014, 426, 1947−1957.(11) Woods, R. J.; Xie, M. H.; Von Kreudenstein, T. S.; Ng, G. Y.;Dixit, S. B. mAbs 2013, 5, 711−722.(12) Yin, Y.; Han, G.; Zhou, J.; Dillon, M.; McCarty, L.; Gavino, L.;Ellerman, D.; Spiess, C.; Sandoval, W.; Carter, P. J. mAbs 2016, 8,1467−1476.(13) Strop, P.; Ho, W. H.; Boustany, L. M.; Abdiche, Y. N.;Lindquist, K. C.; Farias, S. E.; Rickert, M.; Appah, C. T.; Pascua, E.;Radcliffe, T.; Sutton, J.; Chaparro-Riggers, J.; Chen, W.; Casas, M. G.;Chin, S. M.; Wong, O. K.; Liu, S. H.; Vergara, G.; Shelton, D.; Rajpal,A.; et al. J. Mol. Biol. 2012, 420, 204−219.(14) Wei, H.; Mo, J.; Tao, L.; Russell, R. J.; Tymiak, A. A.; Chen, G.;Iacob, R. E.; Engen, J. R. Drug Discovery Today 2014, 19, 95−102.(15) Houde, D.; Arndt, J.; Domeier, W.; Berkowitz, S.; Engen, J. R.Anal. Chem. 2009, 81, 5966.(16) Fang, P.; Zhang, H. M.; Shapiro, R.; Marshall, A. G.; Schimmel,P.; Yang, X. L.; Guo, M. Proc. Natl. Acad. Sci. U. S. A. 2011, 108,8239−8244.(17) Xu, X.; Shi, Y.; Zhang, H. M.; Swindell, E. C.; Marshall, A. G.;Guo, M.; Kishi, S.; Yang, X. L. Nat. Commun. 2012, 3, 681.(18) Zhang, Q.; Willison, L. N.; Tripathi, P.; Sathe, S. K.; Roux, K.H.; Emmett, M. R.; Blakney, G. T.; Zhang, H. M.; Marshall, A. G. Anal.Chem. 2011, 83, 7129−7136.(19) Zhang, A.; Hu, P.; MacGregor, P.; Xue, Y.; Fan, H.; Suchecki, P.;Olszewski, L.; Liu, A. Anal. Chem. 2014, 86, 3468−3475.(20) Houde, D.; Peng, Y.; Berkowitz, S. A.; Engen, J. R. Mol. Cell.Proteomics 2010, 9, 1716−1728.

(21) Zhang, A.; Singh, S. K.; Shirts, M. R.; Kumar, S.; Fernandez, E. J.Pharm. Res. 2012, 29, 236−250.(22) Yu, D.; Song, Y.; Huang, R. Y.; Swanson, R. K.; Tan, Z.;Schutsky, E.; Lewandowski, A.; Chen, G.; Li, Z. J. J. Chromatogr A2016, 1457, 66−75.(23) Walters, B. T.; Jensen, P. F.; Larraillet, V.; Lin, K.; Patapoff, T.;Schlothauer, T.; Rand, K. D.; Zhang, J. J. Biol. Chem. 2016, 291, 1817−1825.(24) Pan, L. Y.; Salas-Solano, O.; Valliere-Douglass, J. F. Anal. Chem.2014, 86, 2657−2664.(25) Zhang, L.; Ren, G. J. Phys. Chem. Biophys 2012, 2, No. e103,DOI: 10.4172/2161-0398.1000e103.(26) Kan, Z. Y.; Mayne, L.; Chetty, P. S.; Englander, S. W. J. Am. Soc.Mass Spectrom. 2011, 22, 1906−1915.(27) Wales, B. T. Ph.D. Thesis, University of Pennsylvania, ProQuestDissertation Publishing, 2013; pp 71−79.(28) Hall, Z.; Robinson, C. V. J. Am. Soc. Mass Spectrom. 2012, 23,1161−1168.(29) Smith, D. L.; Deng, Y.; Zhang, Z. J. Mass Spectrom. 1997, 32,135−146.(30) Cravello, L.; Lascoux, D.; Forest, E. Rapid Commun. MassSpectrom. 2003, 17, 2387−2393.(31) Zhang, H. M.; Kazazic, S.; Schaub, T. M.; Tipton, J. D.; Emmett,M. R.; Marshall, A. G. Anal. Chem. 2008, 80, 9034−9041.(32) Weis, D. D.; Wales, T. E.; Engen, J. R.; Hotchko, M.; Ten Eyck,L. F. J. Am. Soc. Mass Spectrom. 2006, 17, 1498−1509.(33) Zhang, H. M.; Yu, X.; Greig, M. J.; Gajiwala, K. S.; Wu, J. C.;Diehl, W.; Lunney, E. A.; Emmett, M. R.; Marshall, A. G. Proteinscience: a publication of the Protein Society 2010, 19, 703−715.(34) Schlothauer, T.; Rueger, P.; Stracke, J. O.; Hertenberger, H.;Fingas, F.; Kling, L.; Emrich, T.; Drabner, G.; Seeber, S.; Auer, J.;Koch, S.; Papadimitriou, A. mAbs 2013, 5, 576−586.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.7b03830Anal. Chem. 2017, 89, 13494−13501

13501

S-1

Supporting Information

Structural and Functional Characterization of a Hole-Hole

Homodimer Variant in a “Knob-Into-Hole” Bispecific

Antibody

Hui-Min Zhang,*,† Charlene Li,† Ming Lei,† Victor Lundin,† Ho Young Lee,‡ Milady

Ninonuevo,† Kevin Lin, ǀ Guanghui Han,§ Wendy Sandoval,§ Dongsheng Lei,|| Gary Ren,||

Jennifer Zhang,† and Hongbin Liu*,†ǂ

† Protein Analytical Chemistry, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States

‡ Biological Technologies, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, United States

ǀ Analytical Operations, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, United States

§ Departments of Microchemistry, Proteomics and Lipidomics, 1 DNA Way, South San Francisco, CA 94080, United States

|| The Molecular Foundry, Lawrence Berkeley National Laboratory, One Cyclotron Road, Berkeley, CA 94720, United States

ǂ Current address: Nektar Therapeutics, 455 Mission Bay Blvd. South, San Francisco, CA 94158

* Corresponding Authors. H.-M. Zhang E-mail: zhang.huimin@gene.com Tel: 650-467-9618,

Fax: 650-225-3554 or H. Liu E-mail: hliu@nektar.com, Tel: 415-482-5423, Fax: 415-339-5381

This supporting information provides a detailed description of the materials and methods used

for size-exclusion chromatography, native MS, negative-staining electron microscopy, the

antigen binding assay to measure the binding of hole HD with the antigen peptide, AND.

Contents

Size Exclusion Chromatography ................................................................................................... 2

Intact Mass Analysis by Native Mass Spectrometry ..................................................................... 2

Negative-Staining Electron Microscopy (NS EM) ......................................................................... 2

Antigen Binding Assay to Measure the Binding of Hole Homodimer with the Antigen Peptide .... 3

FcRn Affinity Chromatography to Evaluate the Hole-hole Homodimer’s Fc Structural Integrity ... 4

Figure S-1. Hydrophobic interaction chromatography (HIC) profile of the hole-hole homodimers 4

Figure S-2. SEC for the hole-hole homodimer (HD) stored under different conditions and lengths

of time (The insert is the overlay). ................................................................................................. 5

S-2

Figure S-3. FcRn Affinity Chromatography for the hole-hole homodimer (HD) stored at -70 °C

and 2-8 °C, as well as the BsAb1 samples. .................................................................................. 6

REFERENCES ............................................................................................................................. 6

Size Exclusion Chromatography

SEC was performed on an Agilent 1200 HPLC (Agilent Technologies, Santa Clara, CA) with a TSK G3000SWXL column, 7.8 × 300 mm (Tosoh Biosciences, King of Prussia, PA). The mobile phase was 0.2 M potassium phosphate buffer (pH 6.2) containing 0.25 M potassium chloride. Each sample was diluted to 1 mg/mL with the mobile phase. The protein load for each injection was 50 μg. The separation was conducted at ambient temperature with a flow rate of 0.5 mL/min. The column effluent was monitored at 280 nm UV wavelength.

Intact Mass Analysis by Native Mass Spectrometry

For the native MS experiment, 25–50 µg of the hole-hole homodimer samples were buffer-exchanged using Micro Bio-Spin TM 6 columns, prepackaged in Tris buffer (Bio-Rad Laboratories, Inc., Hercules, CA). The column was first centrifuged at 3300 rpm at 4 °C to flush the Tris buffer and then rinsed five times in 100 mM ammonium acetate (Sigma-Aldrich Corp., St. Louis, MO) by loading the column with 0.5 mL of buffer and centrifuging for 1 min at 3300 rpm at 4 °C. The collection tube was emptied after each spin. The column was then placed in a new collection Eppendorf tube, and 25–50 µg of sample were added on the resin at the center of the column. The column was then centrifuged for 10 min at 2000 rpm at 4 °C. The buffer-exchanged sample was collected in the Eppendorf tubes.

Samples were directly infused into an Exactive Plus extended mass range (EMR) Orbitrap mass spectrometer (Thermo Scientific, San Jose, CA) via nanospray ionization using a Triversa TM Nanomate (Advion, Inc., Ithaca, NY). The instrument was set in EMR MS mode for intact mass analysis. The IgG samples were analyzed under the following acquisition parameters: spray gas pressure, 1.0 psi; spray voltage, 1.50 kV; Capillary temperature, 250 °C; S-lens RF level, 100; Scan rage, 1000 to 10000 m/z; Desolvation, in-source CID 120 ev, collision energy, 0; Resolution, 17500 at m/z 200; Polarity, positive; Microscans, 10; AGC targe, 3e6; Maximum injection time, 50; Averaging, 100; Source DC offset, 25V; Injection Flatapole DC, 8V; Inter Flatapole lens, 7V; Bent Flatapole DC, 6V; Transfer multipole DC tune offset, 0V; C-Trap entrance lens tune offset, 0V; Trapping gas pressure setting, 5. Spectra were visualized using Thermo Xcalibur Qual Browser then annotated manually. Mass spectrometric data was analyzed using Protein Deconvolution v4.0 software (Thermo Scientific).

Negative-Staining Electron Microscopy (NS EM)

Conventional cryo-electron microscopy (cryo-EM) is often the method of choice for studies of protein structure under physiological conditions because it avoids potential artifacts from fixatives and stains.1 Still, cryo-EM studies of antibodies are challenging due to their small molecular mass (~150 kDa) and flexible structure.2 Thus, we used an

S-3

optimized negative-staining (OpNS) protocol,1,3 which was refined from the conventional negative staining protocol by using cryo-EM images as a control. OpNS EM images present a much higher image contrast than cryo-EM images with a reasonable resolution (1–3 nm) for visualizing the domains of each antibody. OpNS has been used to examine many proteins, such as antibodies,4,5 GroEL,6 and Glycyl-tRNA synthetase.7 Through those studies OpNS has been validated as a general protocol.

The negative-staining specimens of antibody homodimers were prepared by using optimized negative-staining (OpNS) protocol as described previously.1,3 Briefly, antibody samples were diluted to ~ 0.04 μg mL-1 with Dulbecco's phosphate buffered saline (DPBS). An aliquot (~ 4 μL) of diluted sample was then placed on an ultra-thin carbon-coated 200-mesh copper grid (CF200-Cu-UL, Electron Microscopy Sciences, Hatfield, PA, USA) that had been glow-discharged for 15 s. After 1 min incubation, the excess solution on grid was blotted with filter paper. The grid was then washed with water and stained with 1% (w/v) uranyl formate (pH ~4.5) before air-drying with nitrogen.1,3

Samples were examined by using a Zeiss Libra 120 Plus TEM (Carl Zeiss NTS) operated at 120 kV high tension. The micrographs were acquired under defocus between ~0.6 μm and ~0.9 μm. A dose of ~40-90 e-/Å2 using a Gatan UltraScan 4K X 4K CCD under a magnification of 80 kx (each pixel of the micrographs corresponds to 1.48 Å in specimens) was used. The contrast transfer function (CTF) of each micrograph was examined by ctffind38 and corrected by use of the SPIDER9 software after the X-ray speckles were removed. Particles were then selected from the micrograph with a box size of 256 × 256 by use of boxer (EMAN10 software). All particles were masked by a round mask generated from SPIDER after a Gaussian high-pass filtering. The reference-free class averages of particles were obtained by using refine2d (EMAN software) based on 2631 particles for the antibody sample stored at -70˚C.

Crystal structure of mouse IgG2 antibody (PDB entry 1IGT11) was superimposed on representative images of particles (Y- and X-shapes) to reflect the structural dynamics of the hole-hole homodimer. The two Fab domains and two chains of Fc domains within 1IGT were rigid-body rotated and translated to obtain their best superimposition onto the representative particle images using Chimera.12

Antigen Binding Assay to Measure the Binding of Hole Homodimer with the Antigen Peptide

A 96-well high binding polystylen microtiter plate was incubated for 16-72 h at 4 °C with 100 μL of 3 μg/mL of Neutravidin in DPBS (-Ca2+, -Mg2+). The plate was washed with wash buffer (1x PBS, pH 7.4, 0.05% Polysorbate 20) twice with auto program (6 times wash) using a BioTek 405 plate washer. The plate was blocked with 200 μL of blocking buffer (1x PBS, pH 7.4, 0.05% Polysorbate 20, 0.5% BSA) at 25 °C for 1-2 h. Then the plate was washed as above. 100 μL of biotinylated antigen peptide (2 μg/mL) in assay diluent (1x PBS, 0.05% Polysorbate 20, 0.5% BSA) was added to each well, and incubated at 25 °C for 1 h. While incubating the plate, experimental samples were diluted accordingly using assay diluent. After incubation with the antigen peptide, the plate was washed as described above, and 100 μL of the sample diluents were added to the plate. The sample diluents were incubated at 25 °C for 1 h. After washing, 100 μL of 3 ng/mL of goat anti-human-IgG (Fab’)2 HRP (Jackson ImmunoResearch

S-4

Cat.No.109-036-097) in assay diluent was added and incubated at 25 °C for 1 h. The plate was washed after the incubation, and 100 μL of TMB substrate (SureBlue Reserve TMB Microwell Poroxidase substrate, KPL, cat. No. 53-00) was added and incubated until the color developed adequately. 100 μL of 0.6N sulfuric was added to quench the reaction. The plate was measured for the optical density (OD) values on the plate reader using two wavelengths, 450 nm for detection absorbance and 650 nm for reference absorbance. The data were analyzed with SoftMaxPro software by 4P analysis, and relative potency was calculated using the -70 °C hole homodimer as standard.

FcRn Affinity Chromatography to Evaluate the Hole-hole Homodimer’s Fc Structural Integrity

FcRn affinity chromatography was performed on Thermo/Dionex Ultimate 3000 UHPLC system. The soluble extracellular domain of FcRn with His-Avi-Tag® associated with beta-2-microglubulin was immobilized onto POROS® streptavidin beads that were packed into a 300 μL column13. Mobile phase A was 20 mM MES(2-(N-morpholino)ethanesulfonic acid , 150 mM NaCl, pH 6.0, mobile phase B was 20 mM MES, 150 mM NaCl, pH 6.5 and mobile phase C was 20 mM Tris/HCl, 150 mM NaCl, pH 8.5. The flow rate was 0.25 mL/min. Samples were diluted to 1.0 mg/mL in mobile phase A. Approximately 30 μg of each sample was loaded onto the column, where it was eluted using the following pH gradient: 100% mobile phase A to 66% mobile phase B in 5 minutes; to 75 % B and 25% C in 5 minutes; to 15% B and 75% C in 20 minutes; to 100% C in 8 minutes; hold 100% C for 9 minutes; to 100% A in 2 minutes; hold 100% A for 12 minutes. The elution was monitored by UV absorption at 280 nm, and the pH was monitored using an on-line pH meter.

Figure S-1. Hydrophobic interaction chromatography (HIC) profile of the hole-hole homodimers

Hole-hole homodimer (HD) at different time points during storage at 2-8 °C, after buffer exchange to pH 7.5 from the first buffer exchange (pH 5.8 to pH 3.0) for 1.5, 3 and 18 h, compared with the hole-hole HD frozen control. (Note: the -70 °C controls were made from different batches, so the frozen control here is different from the ones in Figure 1.)

S-5

Figure S-2. SEC for the hole-hole homodimer (HD) stored under different conditions and lengths of time (The insert is the overlay).

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0

-5.0

0.0

10.0

20.0

30.0

40.0

Hole-hole HD Control @ -70 °C

pH 3.0 -> 7.5 – T0

pH 3.0 -> 7.5 – T 1.5 hr

pH 3.0 -> 7.5 – T 3 hr

pH 3.0 -> 7.5 – T 18 hr

Retention Time (min)

Ab

so

rba

nc

e @

28

0 n

m (

mA

U)

0.0 5.0 10.0 15.0 20.0 25.0 30.0-50

0

100

200

300

400

Hole-hole HD (~ 6 mon 2-8 °C @ pH 5.8)

Hole-hole HD (- 70 °C @ pH 5.8)

Monomer HMWS LMWS

Hole-hole HD (2 week 2-8 °C @ pH 5.8)Ab

so

rba

nc

e @

28

0 n

m (

mA

U)

Retention Time (min)

S-6

Figure S-3. FcRn Affinity Chromatography for the hole-hole homodimer (HD) stored at -70 °C and 2-8 °C, as well as the BsAb1 samples.

The hole-hole HD stored at -70 °C displayed three peaks, a flow-through peak (no

binding), a low-affinity binding peak, and a main peak. The hole-hole HD stored at 2-

8 °C displayed a main peak and a small peak with low-affinity binding. The BsAb1

displayed a single main peak. Note that the retention time is slightly different between

the bispecific antibody and the hole-hole HD at 2-8 °C, possibly due to their slightly

different Fc structure.

REFERENCES

(1) Zhang, L.; Song, J.; Newhouse, Y.; Zhang, S.; Weisgraber, K. H.; Ren, G. Journal of lipid research 2010, 51, 1228-1236. (2) Ohi, M.; Li, Y.; Cheng, Y.; Walz, T. Biol Proced Online 2004, 6, 23-34. (3) Zhang, L.; Song, J.; Cavigiolio, G.; Ishida, B. Y.; Zhang, S.; Kane, J. P.; Weisgraber, K. H.; Oda, M. N.; Rye, K. A.; Pownall, H. J.; Ren, G. Journal of lipid research 2011, 52, 175-184. (4) Zhang, X.; Zhang, L.; Tong, H.; Peng, B.; Rames, M. J.; Zhang, S.; Ren, G. Sci Rep 2015, 5, 9803.

BsAb1

0.0 10.0 20.0 30.0 40.0 50.0 60.0 66.0-20

0

50

100

140

Hole-hole HD -70 C

-5.0

0.0

10.0

20.0

30.0

35.0

-10.0

0.0

20.0

40.0

60.0

Hole-hole HD 2-8 C

Ab

so

rba

nc

e@

280 n

m (

mA

U)

Retention Time (min)

Flow-through peak

(no binding) Low-affinity peak

Main peak

S-7

(5) Zhang, L.; Tong, H.; Garewal, M.; Ren, G. Biochim Biophys Acta 2013, 1830, 2150-2159. (6) Rames, M.; Yu, Y.; Ren, G. J. Vis. Exp. 2014, e51087. (7) Deng, X.; Qin, X.; Chen, L.; Jia, Q.; Zhang, Y.; Zhang, Z.; Lei, D.; Ren, G.; Zhou, Z.; Wang, Z.; Li, Q.; Xie, W. J Biol Chem 2016, 291, 5740-5752. (8) Mindell, J. A.; Grigorieff, N. Journal of structural biology 2003, 142, 334-347. (9) Frank, J.; Radermacher, M.; Penczek, P.; Zhu, J.; Li, Y.; Ladjadj, M.; Leith, A. Journal of structural biology 1996, 116, 190-199. (10) Ludtke, S. J.; Baldwin, P. R.; Chiu, W. Journal of structural biology 1999, 128, 82-97. (11) Harris, L. J.; Larson, S. B.; Hasel, K. W.; McPherson, A. Biochemistry-Us 1997, 36, 1581-1597. (12) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J Comput Chem 2004, 25, 1605-1612. (13) Schlothauer, T.; Rueger, P.; Stracke, J. O.; Hertenberger, H.; Fingas, F.; Kling, L.; Emrich, T.; Drabner, G.; Seeber, S.; Auer, J.; Koch, S.; Papadimitriou, A. mAbs 2013, 5, 576-586.